Association of Inherited Pathogenic Variants in Checkpoint Kinase 2 (CHEK2) With Susceptibility to Testicular Germ Cell Tumors

Key Points

Question  Do inherited pathogenic variants in DNA repair genes confer higher susceptibility to testicular germ cell tumors?

Findings  In this multistage case-control study involving 884 men, carriers of germline pathogenic variants in CHEK2 were 4 to 6 times more likely to develop testicular germ cell tumors and, on average, had a 6-year earlier age of presentation than carriers of the wild-type CHEK2 alleles.

Meaning  Inherited CHEK2 mutations are high-risk drivers of susceptibility to testicular germ cell tumors and might be informative for the clinical cancer-risk management of mutation carriers and their at-risk family members.

Importance  Approximately 50% of the risk for the development of testicular germ cell tumors (TGCTs) is estimated to be heritable, but no mendelian TGCT predisposition genes have yet been identified. It is hypothesized that inherited pathogenic DNA repair gene (DRG) alterations may drive susceptibility to TGCTs.

Objective  To systematically evaluate the enrichment of germline pathogenic variants in the mendelian cancer predisposition DRGs in patients with TGCTs vs healthy controls.

Design, Setting, and Participants  A case-control enrichment analysis was performed from January 2016 to May 2018 to screen for 48 DRGs in 205 unselected men with TGCT and 27 173 ancestry-matched cancer-free individuals from the Exome Aggregation Consortium cohort in the discovery stage. Significant findings were selectively replicated in independent cohorts of 448 unselected men with TGCTs and 442 population-matched controls, as well as 231 high-risk men with TGCTs and 3090 ancestry-matched controls. Statistical analysis took place from January to May 2018.

Main Outcomes and Measures  Gene-level enrichment analysis of germline pathogenic variants in individuals with TGCTs relative to cancer-free controls.

Results  Among 205 unselected men with TGCTs (mean [SD] age, 33.04 [9.67] years), 22 pathogenic germline DRG variants, one-third of which were in CHEK2 (OMIM 604373), were identified in 20 men (9.8%; 95% CI, 6.1%-14.7%). Unselected men with TGCTs were approximately 4 times more likely to carry germline loss-of-function CHEK2 variants compared with cancer-free individuals from the Exome Aggregation Consortium cohort (odds ratio [OR], 3.87; 95% CI, 1.65-8.86; nominal P = .006; q = 0.018). Similar enrichment was also seen in an independent cohort of 448 unselected Croatian men with TGCTs (mean [SD] age, 31.98 [8.11] years) vs 442 unselected Croatian men without TGCTs (at least 50 years of age at time of sample collection) (OR, >1.4; P = .03) and 231 high-risk men with TGCTs (mean [SD] age, 31.54 [9.24] years) vs 3090 men (all older than 50 years) from the Penn Medicine Biobank (OR, 6.30; 95% CI, 2.34-17.31; P = .001). The low-penetrance CHEK2 variant (p.Ile157Thr) was found to be a Croatian founder TGCT risk variant (OR, 3.93; 95% CI, 1.53-9.95; P = .002). Individuals with the pathogenic CHEK2 loss-of-function variants developed TGCTs 6 years earlier than individuals with CHEK2 wild-type alleles (5.95 years; 95% CI, 1.48-10.42; P = .009).

Conclusions and Relevance  This multicenter case-control analysis of men with or without TGCTs provides evidence for CHEK2 as a novel moderate-penetrance TGCT susceptibility gene, with potential clinical utility. In addition to highlighting DNA-repair deficiency as a potential mechanism driving TGCT susceptibility, this analysis also provides new avenues to explore management strategies and biological investigations for high-risk individuals.

Testicular germ cell tumors (TGCTs) are the most common malignant neoplasms in young males.^1,2 Family and twin studies have shown that TGCTs have a strong inherited genetic component with an estimated heritability of 48.9%, making TGCTs far more heritable than breast, ovarian, and colorectal cancers.³ Despite this high heritability and familial tendency, to our knowledge, no highly penetrant mendelian TGCT predisposition genes have been identified.⁴ Moderate-risk and low-risk single-nucleotide polymorphisms, identified through genome-wide association studies, are thought to explain only 37% of father-to-son familial risk, leaving most TGCT heritability yet to be elucidated.^3,5-7

A previous study explored the origins of TGCTs by conducting an integrative analysis of tumor genomic and transcriptomic data, showing that TGCTs were exceptionally enriched for arm-level and chromosome-level gains of 1 parental allele with simultaneous loss of the other parental allele, leading to reciprocal loss of heterozygosity.⁸ This DNA double-strand break–enriched genomic signature suggested that increased DNA damage with deficient repair might be central to the development and progression of TGCTs. This notion is further supported by genome-wide association studies that highlighted several validated TGCT-risk single-nucleotide polymorphisms located in or near DNA-damage sensing and repair genes, such as RAD51C and RFWD3.⁹ However, the role of inherited DNA repair defects in TGCT pathogenesis is largely uncharacterized.

We hypothesized that germline pathogenic variants in DNA repair genes (DRGs) that have been associated with susceptibility to other cancer types¹⁰ might be associated with a predisposition to TGCTs. To evaluate this hypothesis, we conducted a multistage case-control analysis to systematically evaluate germline pathogenic variants in 48 Mendelian cancer-predisposition DRGs in unselected 205 men with TGCT compared with 27 173 cancer-free non-Finnish European individuals from the Exome Aggregation Consortium (ExAC) cohort. Significant findings from the discovery phase were replicated in independent cohorts of 448 unselected men with TGCTs and 442 population-matched controls, as well as 231 high-risk men with TGCTs and 3090 cancer-free men.

A total of 205 individuals with TGCT were included in the discovery cohort (eFigure 1 in Supplement 1) from 2012 to 2017. These men were not specifically selected for early-onset disease or a positive family history of TGCTs and are hereafter referred to as unselected. In addition, we used the cancer-free non-Finnish European individuals of the ExAC cohort, excluding the TCGA cases, as a discovery control group (n = 27 173; hereafter referred to as the ExAC NFE cohort).¹¹ In the validation phase, germline data of independent cohorts of 448 unselected Croatian men with TGCTs (hereafter referred to as the Croatian cohort) and 442 population-matched cancer-free men were examined. Germline data of another 231 men with TGCTs who came from families with 2 or more individuals with TGCTs or those with bilateral TGCTs (hereafter collectively referred to as the high-risk TGCT cohort) and an ancestry-matched group of 3090 cancer-free men ascertained from the Penn Medicine Biobank were also examined (eFigure 1 in Supplement 1). This study was approved by the participating institutions (Dana-Farber Cancer Institute, National Cancer Institute, and University of Pennsylvania) (eAppendix in Supplement 1). All individuals in the discovery, validation, and high-risk cohorts provided written consent to institutional review board–approved protocols that allowed for comprehensive genetic analysis of germline samples. This study conforms to the Declaration of Helsinki.¹²

In the discovery phase, principal component analysis was first performed on high-quality nonlinked germline genotypes of 239 individuals with TGCT and a reference group of 938 individuals from the 1000 Genomes Cohort (eAppendix and eFigures 2-4 in Supplement 1).

Forty-eight DRGs with established cancer risk, from the curated COSMIC (Catalogue of Somatic Mutations in Cancer) germline cancer census gene set (v86; http://cancer.sanger.ac.uk/census), were examined in the discovery cohort (eTable 1 in Supplement 1). Genes with significant enrichment of germline pathogenic variants in the discovery cohort were then exclusively explored in the Croatian and high-risk TGCT cohorts.

Germline variants in men with or without TGCTs from all cohorts were independently evaluated for pathogenicity by 3 clinical geneticists (S.H.A., L.C.P., and D.R.S.) using the American College of Medical Genetics and Genomics guidelines.¹³ Only pathogenic and likely pathogenic variants were included in this study (hereafter collectively referred to as pathogenic germline variants). Variants of unknown significance were excluded from all analyses (eAppendix in Supplement 1).

Statistical analysis took place from January to May 2018. Two-sided Fisher exact and binomial tests were used to calculate the odds ratios, 95% CIs, and P values of germline pathogenic variant enrichment in affected vs unaffected cohorts for each of the examined DRGs. P < .05 was considered significant. Top candidate genes, with a false discovery rate of less than 0.05 in the discovery cohort, were selectively evaluated in independent TGCT cohorts. Regression models were used to examine the association between the germline mutational status and the clinical characteristics of individuals with TGCTs (eAppendix in Supplement 1).

The demographic characteristics of all individuals with TGCTs in our study are summarized in eTable 2 and eFigure 5 in Supplement 1 and the eTable in Supplement 2. The exome-wide median target coverage for germline whole-exome sequencing for the discovery cohort samples was 67.24X (interquartile range, 51.97-77.92) (eFigure 6A in Supplement 1). The median sequencing depth of coverage for the examined DRGs (n = 48) was 133.5X (interquartile range, 119.7-151.6) in the discovery TGCT cohort and 59.63X (interquartile range, 54.45-65.36) in the ExAC cohort (eTable 3 in Supplement 1). All samples had satisfactory exome-wide variant call rates, variant transition-to-transversion rates, and genotype quality (eFigures 6B-F in Supplement 1).

In the discovery cohort (n = 205), 20 men with TGCTs (9.8%; 95% CI, 6.1%-14.7%; binomial exact) had germline pathogenic variants (Figure 1A). Of these, 18 individuals (8.8%) harbored heterozygous pathogenic alterations in BARD1 (OMIM 601593) (1 [0.5%]), BRCA1 (OMIM 113705) (1 [0.5%]), CHEK2 (OMIM 604373) (6 [2.9%]), ERCC2 (OMIM 126340) (1 [0.5%]), FANCC (OMIM 613899) (2 [1.0%]), FANCD2 (OMIM 613984) (1 [0.5%]), FANCM (OMIM 609644) (3 [1.5%]), MUTYH (OMIM 604933) (2 [1.0%]), and PMS2 (OMIM 600259) (1 [0.5%]) (Figure 1B). In addition, 2 men with TGCTs (1.0%) carried 1 heterozygous pathogenic variant in CHEK2 plus a second pathogenic variant in NBN (OMIM 602667) or FANCD2 (eTable 4 in Supplement 1).

For each of the examined DRGs (n = 48), we compared the frequency of pathogenic germline variants in 205 European men with TGCTs from the discovery cohort with the frequency of these genetic alterations in 27 173 cancer-free European controls. Using a false discovery rate q value of less than 0.05 for significant associations, our case-cohort analysis of the TGCT discovery cohort showed significant pathogenic variant enrichment in CHEK2 (Figure 1C and Table).

Approximately one-third (8 [36.4%]) of the 22 detected pathogenic DRG variants were in CHEK2, the most commonly mutated gene in the discovery TGCT cohort (Figure 1D). Among 8 men with TGCTs (3.9%) with pathogenic CHEK2 germline variants, 6 (2.9%) carried variants known to cause complete loss of function (LOF) either by protein truncation (n = 5) or by amino acid substitution in a critical functional domain leading to abolished CHEK2 kinase activity (n = 1).^14,15 In addition, 2 men with TGCTs (1.0%) were heterozygous carriers of the low-penetrance variant CHEK2 p.Ile157Thr (c.470T>C; NM_007194.3) (eFigure 7A and eTable 4 in Supplement 1).^16,17 Given their distinct functional and clinical significance,¹⁸ we analyzed pathogenic CHEK2 LOF variants and the low-penetrance CHEK2 p.Ile157Thr variant separately. Compared with 27 173 controls, individuals in the unselected TGCT discovery cohort were almost 4 times more likely to carry a germline CHEK2 LOF variant (odds ratio [OR], 3.87; 95% CI, 1.65-8.86; nominal P = .006; q = 0.018) (Figure 2A; Table). Our analysis also showed a statistically insignificant enrichment of CHEK2 p.Ile157Thr in the discovery cohort of individuals with TGCTs vs controls (OR, 1.15; 95% CI, 0.20-4.39; false discovery rate of 0.21) (Figure 2B; Table).

To further characterize pathogenic CHEK2 variants in TGCTs, we performed targeted Sanger sequencing and copy number alteration analysis of CHEK2 in an independent cohort of 448 unselected Croatian men with TGCTs and 442 cancer-free Croatian men. Six Croatian men with TGCTs (1.3%; 95% CI, 0.49%-2.89%; binomial exact test) had pathogenic CHEK2 LOF variants (eFigure 7B and eTable 5 in Supplement 1), significantly higher than the frequency of these genetic alterations in Croatian cancer-free controls (OR, >1.4; P = .03; Fisher exact test) (Figure 2A). Another 23 Croatian men with TGCTs (5.1%) carried at least 1 copy of the founder low-penetrance CHEK2 p.Ile157Thr variant (OR, 3.93; 95% CI, 1.53-9.95; P = .002; Fisher exact test) (Figure 2B; eTables 5 and 6 in Supplement 1).

In addition to the unselected TGCT cohorts, we analyzed a third independent cohort of 231 high-risk individuals with TGCTs. Relative to 3090 cancer-free men from the Penn Medicine Biobank, high-risk individuals with TGCTs were approximately 6 times more likely to carry a pathogenic LOF variant in CHEK2 (6 men [2.6%]; 95% CI, 1.0%-5.6%; binomial test and OR, 6.30; 95% CI, 2.34-17.31; P = .001; Fisher exact test) (Figure 2A; eFigures 7C and 7D in Supplement 1). Similar to the discovery cohort, there was a small nonsignificant enrichment of CHEK2 p.Ile157Thr in high-risk individuals with TGCTs relative to population-matched cancer-free controls (3 men [1.3%]; OR, 2.13; 95% CI, 0.53-7.02; P = .19; Fisher exact test) (Figure 2B; eTables 7 and 8 in Supplement 1). A subsequent targeted segregation analysis of 3 families with TGCTs showed evidence of cosegregation of pathogenic CHEK2 LOF variants with the TGCT phenotype (eFigure 8 in Supplement 1).

Given the longstanding observation of variable TGCT incidence in genetically distinct populations, with a notable deficit of this disease in Africans and Asians,¹⁹ we next examined the prevalence of germline CHEK2 LOF variants in 53 105 cancer-free individuals from 7 continental populations. Our analysis showed a significantly higher prevalence of CHEK2 LOF variants in European individuals relative to individuals from African or Asian populations, as previously shown in other studies (Figure 3A).¹⁹ Non-Finnish European males, who are 4 to 5 times more likely to develop TGCTs compared with African American males,¹⁹ were also approximately 4 times more likely to carry a CHEK2 LOF variant compared with African Americans (OR, 4.41; 95% CI, 2.21-9.57; P = 4.537 × 10⁻⁷; Fisher exact test). Furthermore, the frequency of the most common CHEK2 LOF variant (c.1100delC) closely paralleled the differential incidence rates of TGCT in low-risk, intermediate-risk, and high-risk populations (Figure 3B), suggesting inherited LOF variants in CHEK2 as a potential monogenic driver for the differential susceptibility to TGCT across genetically distinct populations.

Although CHEK2 was the only gene significantly enriched for pathogenic variants, we evaluated the clinical implications of analyzing a broader clinically actionable DRG panel in men with TGCTs (eAppendix in Supplement 1). Thirteen men with TGCTs (6.3%; 95% CI, 3.4%-10.6%; binomial exact test) carried clinically actionable pathogenic variants, including a highly penetrant alteration in BRCA1 (1 [0.5%]); moderately penetrant alterations in BARD1, CHEK2, and PMS2 (8 [3.9%]); and low-penetrance heterozygous variants in CHEK2 and MUTYH (4 [2.0%]) (eTable 9 in Supplement 1).

The clinical characteristics of individuals with TGCTs in our study, stratified by CHEK2 mutational status, are summarized in eTable 2 in Supplement 1 and the eTable in Supplement 2. Collectively, individuals with CHEK2 LOF variant developed TGCTs approximately 6 years earlier than individuals with TGCT with CHEK2 wild-type alleles (5.95 years; 95% CI, 1.48-10.42 years; P = .009) (Figure 3C). Men with TGCTs who carried CHEK2 p.Ile157Thr also showed a trend, which did not reach statistical significance, toward an earlier age of presentation (1.92 years; 95% CI, −1.37 to 5.22 years; P = .25). Furthermore, individuals who carried the CHEK2 LOF variant appeared to be more likely to develop contralateral TGCTs (OR, 3.70; 95% CI, 1.29-10.59; P = .01) (Figure 3D). Men with TGCTs who carried germline CHEK2 LOF variants were not more likely to have a positive family history of TGCT (OR, 1.86; 95% CI, 0.63-5.51; P = .26), suggesting that TGCTs are a lower-penetrance presentation for the CHEK2-associated cancer predisposition phenotypic spectrum (Figures 3D). The clinical and molecular characteristics of all individuals with TGCTs with pathogenic germline variants in CHEK2 and other DRGs are summarized in Figure 4.

Testicular germ cell tumors are one of the most heritable cancers.³ However, the inherited genomic drivers of TGCT initiation and progression are not fully characterized. Here, we build on our findings of increased DNA double-strand breaks as a unique genomic feature for TGCTs and identify inherited pathogenic DRG variants as potential genomic determinants of TGCT susceptibility in 3 large independent unselected and high-risk TGCT cohorts.

Our systematic discovery case-cohort analysis of deleterious variants in 48 Mendelian cancer-risk DRGs in 205 unselected European men with TGCTs and 27 173 cancer-free European individuals from the ExAC cohort showed a significantly higher than expected rate of inherited CHEK2 pathogenic variants in men with TGCTs. CHEK2 encodes a 543-residue protein kinase (CHEK2) that is phosphorylated by the DNA damage-sensing protein ATM (ataxia-telangiectasia mutated) to subsequently regulate more than 20 downstream effector proteins, including BRCA1, BRCA2, p53, and Rb, which are critical for DNA double-strand break repair,²⁰ cell cycle regulation,²¹ and cellular apoptosis.²² Germline LOF mutations in CHEK2 are known to confer an increased susceptibility to breast, prostate, and colorectal cancers.^23-25 However, no study so far, to our knowledge, has evaluated the prevalence and significance of pathogenic CHEK2 variants, or those affecting other DRGs, in individuals with TGCTs. We have shown that carriers of germline LOF alterations in CHEK2 from all unselected and high-risk cohorts were significantly more likely to develop TGCTs compared with male carriers of the wild-type CHEK2 allele.

In addition, we also delineated the association of CHEK2 p.Ile157Thr in men with TGCTs. CHEK2 p.Ile157Thr is a European founder missense variant that is present in up to 5% of cancer-free individuals from isolated populations in Poland, Finland, and Croatia.²⁶ Functional studies have demonstrated that isoleucine-to-threonine substitution at position 157 impedes CHEK2 homodimerization in a dominant negative manner.²⁷ As a result, CHEK2 p.Ile157Thr results in a modest increase in the risk of breast cancer (OR, 1.6) and colorectal cancer (OR, 1.4).²⁸ Our analysis of 205 unselected and 231 high-risk individuals with TGCTs from European populations, where CHEK2 p.Ile157Thr variants are less common (population frequency, <0.8%), showed modest enrichment of this variant compared with ancestry-specific controls (unselected individuals: OR, 1.15; high-risk individuals: OR, 2.13). However, neither cohort was large enough to be adequately powered to detect such a modest increase in the risk of TGCTs. In the Croatian population, CHEK2 p.Ile157Thr is a common founder variant (population frequency, approximately 1%-2%) that, similar to CHEK2 LOF variants, appears to confer an increased risk of TGCTs. Larger studies in such genetically unique populations are still needed to confirm this observation.

Collectively, our analysis supports germline pathogenic variants in CHEK2 as novel inherited genomic modifiers of susceptibility to TGCT. To our knowledge, CHEK2 is the only Mendelian TGCT predisposition gene that has so far been identified, which carries both mechanistic and clinical importance. Mechanistically, our findings complement a growing body of evidence for CHEK2 as a master regulator of chromosomal segregation, DNA damage repair, and cell cycle regulation during germ cell development and maturation. A study by Bartkova et al²⁹ noted that CHEK2 expression was temporarily down-regulated during the meiotic stages of spermatogenesis, when programmed DNA double-strand breaks are known to occur, suggesting that CHEK2 is essential for DNA damage–induced apoptosis. Other groups have also found that, in contrast to the wild-type oocytes in which apoptosis occurs in response to widespread radiation-induced DNA double-strand breaks, Chek2-deficient mice oocytes escaped apoptosis and accumulated a substantial amount of DNA damage.³⁰ Finally, it has been shown that healthy germ cells lose CHEK2 expression as they progress to invasive TGCTs.²⁹ Our current analysis implied a particular clinical TGCT phenotype in germline CHEK2 LOF carriers, suggesting a potential distinct genomic evolution of TGCTs in the presence of inherited partial CHEK2 deficiency.

Clinically, germline LOF alterations in CHEK2 confer a moderate increase in the risk of TGCTs in approximately 4% of men with TGCT and, at least partially, explain these individuals’ higher risk of TGCTs. Beyond the potential for TGCT risk stratification, the identification of pathogenic germline CHEK2 variants in men with TGCTs could be informative for the clinical management of these individuals. First, men with TGCTs who carry germline CHEK2 LOF variants present significantly earlier than those with sporadic TGCTs. Furthermore, individuals with CHEK2-mutated TGCTs are more likely to develop prostate and colorectal cancers, for which specific cancer screening strategies exist. In addition, as several studies exploring the efficacy of targeted therapeutic interventions (such as CHEK1 and CHEK2 inhibitors) in DNA-repair–proficient and DNA-repair–deficient tumors are under way, our study identified potentially targetable DNA-repair defects that might be exploited in chemotherapy-resistant testicular cancers.

Our study also highlights the robustness of the targeted multistage germline analysis approach to explore novel rare Mendelian cancer predisposition variants that would otherwise be extremely difficult to identify using a single-cohort exome-wide approach.^25,31 A recent large study that explored germline coding variants in 919 men with TGCTs and 1609 cancer-free controls failed to identify major TGCT risk genes.³² Despite the large cohort size, that single-cohort study showed no statistically significant association once corrected for the 1.08 million associations that were conducted. In addition, no other independent cohorts were available to investigate the top nonsignificant candidates. Finally, that study also reported exceptionally low rates of germline CHEK2 LOF variants, rendering the study underpowered to detect genetic risk factors with an OR smaller than 10, as the authors indicated in a recent reanalysis⁴ (eFigure 9 in Supplement 1).

Our study has several limitations. First, the raw genomic and clinical data for the ExAC cohort were not available for analysis, thus limiting our ability to control for potentially relevant variables. This limitation, however, was mitigated by validating our findings in independent cohorts for which genomic data were properly jointly genotyped and analyzed. Second, to achieve a robust population stratification, our study included only individuals with European ancestry. The frequency of these alterations, their penetrance, and the diagnostic yield of genetic testing may, however, vary substantially across genetically distinct populations. Also, insertion and deletion variants were not individually validated, although an orthogonal sequencing method (Sanger sequencing) was used to validate these findings in the replication phase. Finally, the 3 constituent studies were not designed prospectively to address this specific hypothesis, which may introduce unknown biases at the time of a pooled data analysis; hence, larger studies are still needed to further delineate this association.

Our multicenter case-control analysis of 884 individuals with TGCTs and 30 705 controls provides consistent evidence for CHEK2 as a novel Mendelian TGCT susceptibility gene. Inherited pathogenic CHEK2 variants may account for the increased risk of TGCTs in approximately 4% of all unselected individuals with TGCTs, making them potentially informative genetic determinants for TGCT susceptibility and carcinogenesis.

Accepted for Publication: November 13, 2018.

Corresponding Author: Eliezer M. Van Allen, MD, Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, 360 Longwood Avenue, LC-9326, Boston, MA 02215 (eliezerm_vanallen@dfci.harvard.edu).

Published Online: January 24, 2019. doi:10.1001/jamaoncol.2018.6477

Author Contributions: Drs Nathanson and Van Allen had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs AlDubayan and Pyle contributed equally to this work. Drs Nathanson, Van Allen, Stewart, and Lessel jointly supervised work.

Concept and design: AlDubayan, Pyle, Gamulin, Kulis, Greene, Kubisch, Nathanson, Van Allen, Lessel.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: AlDubayan, Pyle, Gamulin, Moore, Hamid, Godse, Jacobs, Meien, Markt, Greene, Sweeney, Kubisch, Nathanson, Van Allen, Stewart, Lessel.

Critical revision of the manuscript for important intellectual content: AlDubayan, Pyle, Gamulin, Kulis, Taylor-Weiner, Reardon, Wubbenhorst, Vaughn, Grgic, Kastelan, Markt, Damrauer, Rader, Kember, Loud, Kanetsky, Greene, Sweeney, Kubisch, Nathanson, Van Allen, Stewart, Lessel.

Statistical analysis: AlDubayan, Pyle, Gamulin, Moore, Taylor-Weiner, Reardon, Wubbenhorst, Markt, Van Allen, Lessel.

Obtained funding: AlDubayan, Gamulin, Rader, Greene, Nathanson, Van Allen, Stewart, Lessel.

Administrative, technical, or material support: Pyle, Gamulin, Kulis, Moore, Godse, Jacobs, Meien, Rader, Loud, Greene, Kubisch, Nathanson, Van Allen, Stewart, Lessel.

Supervision: AlDubayan, Pyle, Gamulin, Kulis, Greene, Sweeney, Kubisch, Nathanson, Van Allen, Stewart, Lessel.

Conflict of Interest Disclosures: Dr Van Allen reported serving as an advisor or consultant for Tango Therapeutics, Genome Medical, Invitae, Illumina, Foresite Capital, and Dynamo; receiving research support from Novartis and BMS; having equity in Tango Therapeutics, Genome Medical, Syapse, and Microsoft; receiving travel reimbursement from Roche/Genentech; and holding institutional patents filed on ERCC2 mutations and chemotherapy response, chromatin mutations and immunotherapy response, and methods for clinical interpretation. Dr Burman received institutional grants for clinical trials from AstraZeneca, Eisai, IBSA, Bayer, and Loxo. Drs Carty, Ferris, McCoy, Seethala, and Yip are employees of University of Pittsburgh Physicians, which is an affiliate of University of Pittsburgh Medical Center (UPMC). Dr Chiosea receives compensation from his employer (UPMC) in connection with ThyroSeq analyses. Dr Mayson previously received research support from Rosetta Genomics. Dr Nikiforov and Dr Nikiforova have intellectual property rights related to ThyroSeq. They will receive royalties associated with commercial use of ThyroSeq. Dr Sipos received an honorarium for speaking for Veracyte and Genzyme. Dr Sosa is a member of the Data Monitoring Committee of the Medullary Thyroid Cancer Registry supported by GlaxoSmithKline, Novo Nordisk, Astra Zeneca, and Eli Lilly. Dr Steward received research funding from Rosetta, Veracyte, and GeneproDX. Dr Handorf reported receiving grants from Pfizer, paid to her institution, outside the scope of the current work. No other disclosures were reported.

Funding/Support: This work was conducted with support from Harvard Catalyst, the Harvard Clinical and Translational Science Center (National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health Award UL1 TR001102) and financial contributions from Harvard University and its affiliated academic healthcare centers. This work was also supported by American Society of Clinical Oncology (ASCO) Conquer Cancer Foundation Career Development Award (CCF CDA) 13167 (Dr AlDubayan), Cancer Genomics Fellowship (KSAU-hs) (Dr AlDubayan), University of Pennsylvania Clinical and Translational Science Award KL2TR001879 (Dr Pyle), grant R01CA222574 from the National Institutes of Health (Dr Van Allen), grant R01CA164947 from the National Institutes of Health (Dr Nathanson), grant R01CA114478 from the National Institutes of Health (Dr Nathanson), Movember Foundation (Dr Van Allen, Dr Nathanson), Damon Runyon Clinical Investigator Award (Dr Van Allen), Forschungsförderungsfonds der Medizinischen Fakultät der Universitätsklinikum Hamburg-Eppendorf (Dr Lessel), Forschungsförderungsfonds der Medizinischen Fakultät der Universitätsklinikum Hamburg-Eppendorf and Deutsche Krebshilfe (Dr Lessel), and the Intramural Research Program of the Division of Cancer Epidemiology and Genetics of the National Cancer Institute as well as the support services contracts HHSN261200655004C and HHSN261201300003C with Westat, Inc (Drs Stewart, Greene, and Loud).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Group Information: The members of the Regeneron Genetics Center Research Team are: Cristopher Van Hout, Colm O’Dushlaine, Alexander Li, Shane McCarthy, Anthony (Tony) Marcketta, Bin Ye, Tanya Teslovich Dostal, Claudia Schurmann, Lauren Gurski, Dylan Sun, Daren Liu, Goncalo Abecasis, Jonathan Marchini, Jeffrey (Jeff) Reid, Ricardo (Rick) Ulloa, Lukas Habegger, John Penn, Xiaodong (Sheldon) Bai, Young Hahn, Jeffrey Staples, Evan Maxwell, Kia Manoochehri, Alicia (Ali) Hawes, Ashish Yadav, Leland Barnard, Gisu Eom, Claire Chai, Li Xu, Shareef Khalid, Manasi Pradhan, Lyndon Mitnaul, Marcus Jones, Melina Bortoli, Paloma Maria Guzzardo, Biming Wu, Aris Baras, Alan Shuldiner, Nehal Gosalia, John Overton, Maria Sotiropoulos-Padilla, Erin Fuller, Alexander (Alex) Lopez, Sarah Wolf, Thomas Schleicher, Michael Lattari, Caitlin Forsythe, Karina Toledo, Louis Widom, Christina Beechert, Zhenhua Gu, Yuri Santos, Claudia Gonzaga-Jauregui, Jan Freudenberg, Suganthi Balasubramanian, Kavita Praveen, Julie Horowitz, Nilanjana (Nila) Banerjee, Sunilbe (Suni) Siceron, and Andrew Blumenfeld. All are affiliated with Regeneron Genetic Center.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic healthcare centers, or the National Institutes of Health.

Additional Contributions: We thank all individuals who participated in this study. We thank Bin Zhu, PhD, and Kristie Jones, BS, at the Cancer Genome Research Laboratory of the Division of Cancer Epidemiology and Genetics of the National Cancer Institute, Gaithersburg, Maryland, for their involvement and supervision of exome sequencing, quality control, and assembly of data.

Additional Information: The UPMC has granted CBLPath Inc a license to market ThyroSeq for commercial use. They receive no compensation, directly or indirectly, related to CBLPath Inc. All Binary Alignment Map (BAM) files of the individuals with TGCT who were examined in the discovery cohort are deposited in the database of Genotypes and Phenotypes (dbGaP) (phs000178.v9.p8 for the Cancer Genome Atlas cohort and phs000923.v1.p1 for the Dana-Farber Cancer Institute cohort). Raw sequencing data for the Institute of Cancer Research (UK) cohort were deposited in the European Genome–Phenome Archive, which is hosted by the European Bioinformatics Institute (accession code: EGAS00001001084).